Can the choice of biologic therapy in psoriasis actually prevent the development of psoriatic arthritis? This is one of the most clinically compelling questions in the psoriatic disease field, and it sits at the intersection of dermatology and rheumatology. 

At EULAR 2026, Dr Nikolaos Kougkas presented data from a unique real-world cohort built within joint dermatology–rheumatology university centers, following 394 patients with psoriasis on bDMARDs for up to 17 years. 

The study compared PsA incidence across bDMARD classes and found a striking signal: patients treated with non-TNFi biologics — whether IL-17, IL-23, or IL-12/23 inhibitors — were significantly less likely to develop PsA than those on TNF inhibitors. 

We sat down with Dr Kougkas to unpack the methodology, challenge the results, and draw out the clinical implications for everyday practice.

Study at a Glance

Q&A

On the Study Design & Cohort

Q1. What drove the decision to set up joint dermatology–rheumatology centers for this cohort, and how did that collaboration shape the quality of your clinical data compared to insurance claims databases?

The collaboration arose naturally from the existence of a combined rheumatology–dermatology clinic, which has been running for approximately six years. The close partnership with dermatologists was essential: it ensured that psoriasis diagnoses were confirmed by dermatologists, and that PsA diagnoses were made by rheumatologists, guaranteeing a dual-specialty validation that insurance claims databases simply cannot replicate. In a claims database, diagnostic codes may be miscoded or overlapping, whereas our real-world cohort benefits from actual clinical assessments at each visit. This is a key methodological advantage that gives us confidence in the robustness of the PsA incidence data.

Q2. How did you handle patients who switched between bDMARD classes during follow-up — was there any sensitivity analysis to address potential exposure misclassification?

We conducted four distinct analytical approaches to address this challenge. The first was based on the first bDMARD received; the second — and primary analysis — classified patients by the bDMARD class used for the longest cumulative duration, which best reflects real-world exposure; the third restricted the analysis to patients who received only a single bDMARD class throughout follow-up; and the fourth was a calendar-period sensitivity analysis including only patients enrolled after 2018, when all bDMARD classes (TNFi, IL-17i, IL-23i, and IL-12/23i) were available in clinical practice. This fourth analysis included 215 patients and yielded consistent results, confirming that the findings are not an artifact of temporal treatment trends or differential availability of biologics.

Q3. Seventeen years is a remarkable follow-up window. How did you manage loss to follow-up, and could attrition have biased which patients ended up in the TNFi group?

Patients who were lost to follow-up were excluded from the analyses to avoid introducing informative censoring. We also adjusted for key potential confounders including methotrexate use and disease duration, which are important covariates given the long timeline. Regarding the TNFi group specifically, it is true that early in the cohort’s history, TNFi agents were the only available bDMARDs, so patients with longer disease duration and potentially more severe disease were preferentially placed on TNFi. However, the consistency of results across all four analytical approaches — including the post-2018 calendar sensitivity analysis — supports the validity of the findings independent of this historical channeling effect.

On the Results

Q4. The PsA incidence in the TNFi group was strikingly high at around 44–45%. Do you think this reflects a channeling bias, where higher-risk patients were preferentially started on TNFi earlier in the cohort?

PsA occurred in 45% of patients treated with TNFi, compared with 14% in the IL-17i group (p<0.001), 10% in the IL-23i group (p<0.001), and 11% in the IL-12/23i group (p=0.002). Channeling bias is certainly a legitimate concern: in the single-bDMARD subgroup analysis, patients who had received only one class of biologic tended to be those enrolled earlier in the cohort, with longer disease duration and potentially more established or severe psoriasis. These are precisely the patients who were placed on TNFi because it was the only available option at the time. Nevertheless, the finding is replicated across all four analytical methods, including in the post-2018 cohort where all classes coexisted, which substantially mitigates this concern.

Q5. No significant differences were found among non-TNFi classes. Does that surprise you, given the mechanistic distinctions between IL-17 and IL-23 inhibition in joint versus skin inflammation?

It is indeed mechanistically interesting. IL-17 and IL-23 act at different points in the inflammatory cascade, and one might hypothesize differential effects on entheseal or synovial tissue. However, our study was not powered to detect differences within non-TNFi classes, and the overall trend was consistent: all three non-TNFi groups showed markedly lower PsA incidence compared to TNFi. Differentiating between IL-17i and IL-23i in terms of PsA interception would require a dedicated, prospective, head-to-head study with arthritis endpoints — a study that does not yet exist.

Q6. How do you interpret the IL-12/23i signal, given that ustekinumab is now largely superseded by more selective agents?

The IL-12/23i group — essentially ustekinumab — was the first non-TNFi option available in our cohort and therefore often constituted an early bridge or first non-TNFi step in patient treatment sequences. Despite not being statistically different from IL-17i or IL-23i in terms of PsA incidence, it clearly performed better than TNFi. One important advantage of our real-world study is that PsA diagnosis was confirmed by rheumatologists, not derived from insurance codes, which strengthens the signal for all agents, including ustekinumab.

On Clinical Implications

Q7. If you had to counsel a dermatologist today on biologic selection for a psoriasis patient with PsA risk factors, what would your take-home message be?

The message is clear: avoid TNFi when possible in psoriasis patients at high risk for PsA. Among the non-TNFi options, IL-23 inhibitors are my preferred first choice — and the earlier you start, the better. The data suggest that intervention before musculoskeletal involvement becomes established is where interception may have its greatest impact. Of course, individual patient factors, comorbidities, and treatment access must be factored in, but from a purely preventive rheumatology standpoint, this study provides real-world evidence to support a proactive biologic strategy using IL-23 inhibitors in high-risk psoriasis patients.

Editorial Perspective

This study is thought-provoking and adds meaningful real-world data to the growing concept of psoriatic disease interception. 

What stands out is not just the magnitude of the difference in PsA rates between TNFi and non-TNFi groups, but the methodological rigor brought by a genuine rheumatology–dermatology partnership — a model that is still the exception rather than the rule. 

The channeling bias question is legitimate and the authors address it honestly; the post-2018 sensitivity analysis is reassuring, though the relatively modest sample size in that subgroup means we should interpret results with appropriate caution. The inability to differentiate between IL-17i and IL-23i on PsA prevention is a limitation, but it also opens the door for the next generation of studies. 

For now, this is exactly the type of real-world signal that should prompt rheumatologists and dermatologists to think more strategically, and collaboratively, about biologic selection in psoriasis patients before joint disease ever begins.